"UWB RF Sensor Phantom Measurements for Brain Tissue Characterization: 47-Configuration S11 Feature Dataset with Edge-Deployed ML"

"This dataset contains 47 tissue-mimicking phantom measurements from NeuroSense, a wearable Ultra-WideBand (UWB) RF sensing platform for non-invasive brain abscess detection. The system integrates two antenna designs \u2014 a circular microstrip patch at 6.0 GHz and an M-shaped slot antenna at 1.55 GHz with 1.1 GHz bandwidth \u2014 both on FR-4 substrate, coupled with a battery-operated LiteVNA64 portable vector network analyser and a 3D-printed wearable headband. Each measurement comprises 10 RF signal features extracted from S11 return-loss data across magnitude, phase, and frequency domains. Phantom configurations vary inclusion diameter (5\u201325 mm), depth (10\u201330 mm), and lateral position to simulate healthy tissue and three abscess stages. A dual-model XGBoost\u2013BiLSTM classifier deployed on Raspberry Pi 5 achieves 97.9% binary accuracy under leave-one-out cross-validation (AUC = 0.996; sensitivity = 1.000; specificity = 0.833; 95% CI: 88.9\u201399.6%). A reproduction script is included. This dataset supports the associated Nature submission."